Protection of geographical indication for Xihu longjing tea was necessary and an accurate method of identification of Xihu longjing tea ought to be developed to prevent Xihu longjing tea from being faked and protect the trade mark of Xihu longjing tea. Identification of Xihu longjing tea with near-infrared spectrum was researched and a new method for identification of Xihu longjing tea by a quantitative model was developed. A new variance was first defined for Xihu longjing tea and other flat-shaped tea. And then the near-infrared spectrum without preprocess of tea was analyzed and partial least squares (PLS)was most suitably used to build the model (The number of PLS factors in this model is 9) for quantitative prediction. Compared with the critical numeral value, the result predicted by the PLS model was used to identify true Xihu longjing tea. The model was used to predict for 70 known samples which were used to build the quantitative model and 24 unknown samples. The recognition rate of 100% was achieved. So it is proved that the quantitative model by PLS was reliable and accurate to identify real Xihu longjing tea.